{"id":"b45c6e7e-4509-44f8-940f-a1bf9e890ed0","arxiv_id":"2607.17400","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A four-wave-plate stack plus an eight-Stokes-vector protocol can compensate two fiber segments around an embedded controller, holding polarization-induced excess QBER below 1% even 118 nm off the design wavelength.","lead":"This paper shows how to build a four-wave-plate polarization compensator that works even when the light wavelength is far from the plates' design value, and how to reconstruct the two fiber segments on either side of the compensator. It matters because quantum key distribution networks need flexible wavelength channels and compensators placed at intermediate nodes, not just at the receiver.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"SO(3) channel assumption is load-bearing: unquantified PDL/depolarization in the DUTs could invalidate the eight-Stokes reconstruction and the compensation formula.","rationale":"The reader identified the same load-bearing premise: every link segment is assumed to be a loss-normalized, non-depolarizing SO(3) channel. I agree that this is the core assumption on which both the reconstruction (Eqs. 4-9) and the compensation condition (Eq. 3) rest. The paper explicitly excludes PDL and depolarization, but the abstract's 'deterministic polarization compensation framework' and the experimental claims could be over-read as applying to arbitrary optical links. The DUTs used in the experiment may well be nearly SO(3), but the paper reports no Mueller polarimetry that would quantify PDL or depolarization, so the validity of the central premise in the demonstrated regime is unverified. The secondary concern about accurate realization of the test rotations N_i is real but less fundamental; it can, in principle, be addressed by better calibration, whereas a non-SO(3) link invalidates the method itself. I do not see a flaw in the mathematical derivations: the Q-Q-Q-H universality proof and the winding-number argument appear internally consistent, and the experimental results are credible though statistically thin. The reader's conditional verdict, asking for raw data and a clearer PDL/depolarization boundary, is exactly the right response. My stress-test does not move that verdict, so I keep it unchanged.","tokens_in":28502,"tokens_out":18939,"duration_ms":185510,"concrete_test":"Perform full 4x4 Mueller polarimetry on each DUT (1 m SMF, 5 m fiber-on-paddles, optical switch, 100 m spool) at the reported wavelengths. Compute the polarization-dependent loss (from the singular values of the Mueller matrix) and the depolarization index. Then apply the eight-Stokes reconstruction to the measured 4x4 data, both with and without projecting onto SO(3), and compare the residual end-to-end QBER after compensation. If the added QBER from the non-SO(3) part is below about 0.1% for all DUTs, the SO(3) premise is validated for the demonstrated regime; if it is larger, the central claim needs a PDL/depolarization qualifier or a PDL-tolerant reconstruction.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that the eight-Stokes protocol reconstructs M_alpha and M_beta separately and that M = M_beta^T M_alpha^T gives deterministic compensation—requires every link segment to be a loss-normalized, non-depolarizing Mueller matrix in SO(3). The paper is explicit: Supplemental S1 after Eq. (S1) states 'Polarization-dependent loss and depolarization are not treated as compensable errors in this Letter.' The reconstruction uses the SO(3) structure at three points: (i) Eq. (4) fills the third column of E_i via a cross product; (ii) Eq. (6) cancels M_alpha using E_0^T, relying on M_alpha^{-1}=M_alpha^T; (iii) Eq. (8) treats the extracted rotation axes u_i as an orthonormal frame for M_beta. If a real segment has PDL or depolarization, the measured and intensity-normalized 3x3 matrix is not a rotation. The orthonormalization in S2.B then projects this non-rotation onto SO(3), biasing M_beta and hence M_alpha and M. The paper never quantifies the PDL or depolarization of its own DUTs (1 m SMF, 5 m fiber paddles, optical switch, 100 m spool), so it is unknown whether the reported sub-percent QBERs occur because the DUTs are close to SO(3) or despite a systematic model violation. A secondary related fragility is that the test rotations N_i are implemented with the same Q-Q-Q-H stack; if the actual N_i deviate from the nominal matrices, the extracted u_i are rotated by the stack's error, again biasing M_beta. The SO(3) premise is the more fundamental issue because it affects every step of the reconstruction and is not tested by the paper's data.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents a deterministic polarization-compensation framework for quantum optical links in which the compensator is embedded between two unknown channel segments. It makes two main contributions. First, it shows that a standard Q-Q-H wave-plate sequence is not universal when the wave-plate retardances deviate from their ideal values, and it introduces a Q-Q-Q-H four-wave-plate stack that can synthesize arbitrary SO(3) Mueller matrices under the sufficient condition |Δδ+|+|Δδ1|+|Δδ2|+|Δδ3| < π/2 (Supplemental S1, Eq. S104). A constructive angle-solving algorithm is provided. Second, it develops an eight-Stokes-vector protocol (Eqs. 4–9) that reconstructs the Mueller matrices Mα and Mβ of the link segments before and after an embedded compensator, allowing the required compensation matrix M = Mβ^T Mα^T to be computed directly. The protocol is validated experimentally with commercial zero-order 633 nm wave plates: at 640 nm and 515 nm the Q-Q-Q-H stack gives sub-percent polarization-induced excess QBER, in contrast to a Q-Q-H baseline, and low residual QBER is reported for various fiber devices and an optical switch. The paper further demonstrates auxiliary-wavelength interpolation and tracking, maintaining median excess QBER of 0.115% during a 14 h temperature-variation experiment on a 100 m fiber spool.","tokens_in":28911,"tokens_out":9386,"duration_ms":96301,"significance":"If the claims hold, the paper solves a genuine and practically relevant problem: compensator placement at intermediate network nodes requires separate knowledge of the two surrounding segments, and wavelength-flexible operation requires compensation schemes robust to non-ideal retardances. The mathematical derivation is detailed and internally consistent, and the experimental results support the central mechanism rather than being obtained by fitting parameters to minimize QBER. The paper is also commendably explicit about its scope, including the restriction to loss-normalized, non-depolarizing channels and the fact that Eq. (S104) is only a sufficient condition. The auxiliary-wavelength interpolation is a heuristic but is tested honestly, including a case where wide spacing fails. The main weakness is that the load-bearing SO(3) assumption and the accuracy of the test rotations are not quantitatively characterized, which leaves the experimental validation somewhat conditional. Still, the ideas are novel and the evidence is substantial for a Letter.","major_comments":[{"comment":"The reconstruction formulas rely on Mα, Mβ ∈ SO(3) at three points: Eq. (4) fills the third column by cross product, Eq. (6) cancels Mα using E0^T = Mα^{-1}, and Eq. (8) treats the extracted axes as an orthonormal frame. The paper explicitly states (Supplemental S1, after Eq. S1) that polarization-dependent loss and depolarization are not treated as compensable errors, but it never quantifies PDL or depolarization for the actual DUTs (1 m SMF, 5 m paddles, optical switch, 100 m spool). If a segment has even modest PDL, the measured intensity-normalized 3×3 matrix is not a rotation, and the SVD projection in S2.B/C will bias Mβ and hence Mα and the computed compensation. The reported sub-percent QBERs therefore do not distinguish a valid SO(3) regime from a hidden model violation. Please add a characterization of the DUTs' non-unitarity (e.g., Mueller polar decomposition or degree-of-pola","section":"Link-segment reconstruction; S2.B/C"},{"comment":"The protocol assumes the four test transformations N0..N3 are known exactly. In the experiment these N_i are implemented with the same Q-Q-Q-H stack, whose actual output rotations inherit the same retardance and positioning uncertainties the method is designed to overcome. Errors in the realized N_i rotate the extracted axes u_i, biasing Mβ; the subsequent orthonormalization removes only non-orthogonality, not a common systematic rotation. No independent calibration of the implemented N_i is reported. This is load-bearing because the reconstruction of Mβ and Mα is the basis for the compensation matrix in Eq. (3). Please quantify the accuracy of the realized N_i (for example, by measuring the actual Stokes response at each setting and comparing with the nominal N_i) or provide an error-budget argument that such errors are negligible at the reported QBER level.","section":"Eq. (4)–(8) and Supplemental S2"},{"comment":"The experimental validation consists of single compensation runs: the stated x±y statistics are the mean and standard deviation over scanned input azimuths, not over repeated trials or independent reconstructions. This makes it difficult to assess run-to-run reproducibility, especially for the 0.051% and 0.055% figures, which are close to the measured hardware baseline of roughly 0.018% (Supplemental S3). Please report the number of independent reconstructions/compensation attempts and the spread of the resulting QBER values, or clearly state that each entry is a single run. This does not affect the mathematical claims, but it strengthens the empirical support.","section":"Table I and four-DUT results"}],"minor_comments":[{"comment":"The text says the input state is prepared by 'a QWP followed by a linear polarizer'. If the polarizer comes after the QWP, the QWP would have no effect on the prepared state; presumably the order is a linear polarizer followed by a QWP. Please correct the wording or the figure labeling.","section":"Fig. 1 and caption"},{"comment":"Minor typo: 'we evaluate that the the sufficient condition is fulfilled' should read 'the sufficient condition'.","section":"Synthesis, main text"},{"comment":"The interpolation uses a linear dependence on 1/λ and SLERP for the rotation axis. This is a reasonable heuristic, but it is not derived from a physical model; the paper should state more explicitly that this is an empirical model validated only on the tested DUTs and that failure for the wide-spacing spool case is expected under this model.","section":"Supplemental S4.C"},{"comment":"Equation (S124) gives the rotation axis extraction for general Θ. The main text and experiment use Θ=π/2; this choice is justified as maximizing the antisymmetric part. It would be helpful to state the expected noise amplification for small deviations of Θ from π/2, since the test rotations are themselves implemented with finite accuracy.","section":"Supplemental S2.A"}],"recommendation":"major_revision","confidential_remarks":"The central mathematical contribution appears sound and the experimental demonstration is genuinely suggestive. My main reservation is that the reconstruction protocol's validity is conditional on the DUTs being loss-normalized, non-depolarizing, and on the test rotations being accurately known; the paper does not yet provide enough quantitative evidence for either premise. These concerns are fixable by adding relatively straightforward characterizations, and I would be prepared to accept after those additions. I do not see grounds for rejection: the limitations are stated rather than concealed, and the core derivation is detailed and self-consistent."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth a careful read. The paper delivers two things: a proof that a generic non-ideal Q-Q-H stack is not universal over SO(3), with an explicit obstruction for rotations about the R-axis, and a four-plate Q-Q-Q-H sequence that restores universality under the retardance-error bound |Δδ+|+|Δδ1|+|Δδ2|+|Δδ3| < π/2. The proof is a winding-number argument: the residual rotation angle around R winds twice as the relative plate angle runs over a 2π period, so it must hit zero. I checked the steps in the supplement; they are coherent. The eight-Stokes reconstruction protocol for a compensator embedded between two unknown link segments is also new as far as the cited literature goes, and it is explicit: four settings of the stack, two input states each, cross-product for the third column, then extraction of Mβ from conjugate rotations and Mα from E0. The two methods fit together cleanly.\n\nThe experiments support the central claims. The 515 nm result (0.41% excess QBER at 118 nm from design wavelength) is a convincing demonstration that the additional QWP helps; the Q-Q-H baseline degrades to ~20% there. The auxiliary-wavelength tracking on the 100 m spool (median 0.115% over 14 h) is a reasonable proof of concept, and the authors are careful to note that the 2 nm spacing test is a scale test for WDM spacing.\n\nNow the soft spots. The validation is statistically thin. Table I and the component results report mean and standard deviation over the scanned input azimuths from what appear to be single runs; there are no repeated trials, and no raw data or code are shipped. The \"data available on request\" line is weak for a paper whose reconstruction procedure is straightforward to benchmark. The stress-test worry about PDL and depolarization is real but the paper is upfront about it: the SO(3) model is assumed, and PDL/depolarization are explicitly not compensable errors. What is missing is any measurement of how close the actual DUTs are to SO(3). For the 1 m fiber and paddles that is probably a non-issue, but the optical switch could have non-negligible PDL, and they never say. A referee should ask for those numbers.\n\nThe theoretical core holds up. The sufficient condition is conservative; the 515 nm operation is outside it and the authors admit the solver was checked numerically, not proved. That is an honest statement, not a flaw.\n\nOverall: a solid engineering paper that reduces two practical constraints in polarization-encoded links. It deserves serious peer review, with a request for more statistics, raw data, and PDL/depolarization characterization of the tested components.","headline":"A genuinely useful engineering result: a four-plate compensator that tolerates wave-plate retardance errors and an eight-Stokes protocol for middle-link compensation, with experiments that support the claims but are statistically thin.","tokens_in":29387,"tokens_out":2224,"would_cite":true,"duration_ms":24716,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["03.67.-a","42.25.Ja","42.81.Gs"],"model":"deepseek-v4-flash","headline":"A four-plate compensator plus an eight-Stokes reconstruction makes middle-of-the-link polarization compensation deterministic and broadband.","keywords":["polarization compensation","SO(3) Mueller matrix","quantum key distribution","wave-plate synthesis","link-segment reconstruction","Stokes vector tomography","auxiliary wavelength tracking","fiber birefringence"],"falsifier":"Insert a known partial polarizer (polarization-dependent loss) between the compensator and the receiver, run the eight-Stokes protocol, and check whether the reconstructed M_beta stays a rotation and whether residual QBER stays below 1%; the paper's model predicts it will not, because PDL is explicitly excluded.","tokens_in":28387,"feed_emoji":"📡","tokens_out":7914,"duration_ms":71534,"temperature":0.7,"pith_summary":"This paper tries to establish that a polarization compensator placed inside a quantum optical link—not just at the receiver—can be made deterministic and broadband. The first claim is a synthesis result: three quarter-wave plates followed by a half-wave plate (Q-Q-Q-H) can produce any rotation of the Poincaré sphere (any SO(3) Mueller matrix) even when the wave plates' retardances deviate from their nominal values, provided the total retardance error is below π/2. The second claim is a measurement result: eight Stokes-vector measurements—two input states for each of four compensator settings—reconstruct the unknown Mueller matrices of the fiber segments before and after the compensator, so the correct compensating rotation can be computed directly. If both claims hold, a compensation node can be placed flexibly inside a network, operated far from the design wavelength, and updated without disturbing the quantum signal. The experiments support this: 0.41% polarization-induced excess QBER at 515 nm (118 nm off the 633 nm design), below 1% during 14 hours of temperature cycling with auxiliary-wavelength feedback, and 0.03–0.18% for several fiber and switch links.","feed_headline":"Four wave plates plus eight Stokes measurements fix both link segments","feed_subtitle":"Sub-percent polarization error at 515 nm—118 nm off design—with no re-optimization.","key_machinery":"The Q-Q-Q-H stack—an extra quarter-wave plate in front of the classic Q-Q-H sequence—preconditions the target rotation so that the remaining three plates remain within a solvable branch; the sufficient condition ensures a continuous winding number that guarantees a zero residual. The eight-Stokes reconstruction uses the identity setting and three π/2 test rotations: for each setting, measuring H and D outputs yields two columns of the end-to-end matrix and the third by cross product; multiplying by the transpose of the identity-setting matrix isolates M_beta Ni M_beta^T, whose rotation axes give M_beta directly, and then M_alpha = M_beta^T E0. For auxiliary-wavelength tracking, measured Muel","core_discovery":"On the paper's own terms, the central discovery is that a compensator embedded between two unknown polarization-transforming link segments can be set without iteration: by measuring eight Stokes vectors (H and D inputs for four compensator settings, N0=I and three π/2 rotations about the Stokes axes), one reconstructs the two segment matrices M_alpha and M_beta, and the required compensator is M = M_beta^T M_alpha^T. The paper also argues that the Q-Q-Q-H wave-plate stack can synthesize any SO(3) matrix under the conservative error bound |Δδ+|+|Δδ1|+|Δδ2|+|Δδ3| < π/2, by using the extra quarter-wave plate to precondition the target and a winding-number argument to guarantee a solution. Exper","pith_inferences":["Because the reconstruction only assumes SO(3) segments, it should transfer to any two-sided optical element whose internal state can be switched among known rotations; the same eight-measurement pattern could serve as a general tomography routine for embedded devices.","Extending the method to links with polarization-dependent loss or depolarization would require full 4×4 Mueller matrices and more than eight measurements; the paper explicitly leaves that case uncompensated.","The π/2 sufficient condition is conservative—the solver succeeded numerically at 515 nm where the condition fails—so the guaranteed wavelength range could probably be widened with a sharper analysis.","The demonstrated auxiliary spacing of 3×10^-3 relative wavelength suggests a telecom implementation at 1550 nm would need about 4.6 nm spacing, i.e., roughly six 100-GHz DWDM channels, which is feasible but leaves fewer intermediate channels."],"forward_implications":["With the Q-Q-Q-H stack, compensation works at 515 nm using 633 nm wave plates, with 0.41% excess QBER; the baseline Q-Q-H sequence gives 19.8% at the same wavelength.","Auxiliary-wavelength feedback kept excess QBER below 1% (median 0.115%) over 14 hours while the fiber spool was cycled between 15°C and 30°C.","The same reconstruction–synthesis procedure gives 0.03–0.18% excess QBER for 1 m fiber, 5 m fiber on paddles, an optical switch, and a 100 m spool.","The demonstrated 3×10^-3 relative auxiliary-wavelength spacing maps to about 4.6 nm at 1550 nm, on the order of six 100-GHz DWDM channels, so auxiliary-wavelength compensation is compatible with dense WDM.","Because the compensator setting is computed from eight measurements rather than iterative search, the method supports placement of the compensator at intermediate network nodes."],"fun_headline_variants":["Eight Stokes vectors set both link segments with no iteration","Four-plate stack plus eight Stokes fixes polarization at 118 nm off-design","Sub-percent QBER from a non-iterative two-segment polarization fix","No re-optimization: four plates track drift and hold QBER below 1%","Non-iterative compensator reconstructs both link segments from eight Stokes"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that every link segment is a loss-normalized, non-depolarizing channel whose Mueller matrix belongs to SO(3); if polarization-dependent loss or depolarization appears, the reconstruction equations and the compensation formula M = M_beta^T M_alpha^T no longer apply.","fun_headline_variants_meta":{"raw":{"variants":["Eight Stokes vectors set both link segments with no iteration","Four-plate stack plus eight Stokes fixes polarization at 118 nm off-design","Sub-percent QBER from a non-iterative two-segment polarization fix","No re-optimization: four plates track drift and hold QBER below 1%","Non-iterative compensator reconstructs both link segments from eight Stokes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000161,"raw_usage":{"total_tokens":1077,"prompt_tokens":755,"completion_tokens":322,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":499,"completion_tokens_details":{"reasoning_tokens":223}},"tokens_in":499,"tokens_out":322,"duration_ms":3635,"temperature":1.0,"reasoning_tokens":223,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T18:03:27.370913+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Insert a known partial polarizer (polarization-dependent loss) between the compensator and the receiver, run the eight-Stokes protocol, and check whether the reconstructed M_beta stays a rotation and whether residual QBER stays below 1%; the paper's model predicts it will not, because PDL is explicitly excluded.","supporting_citations":[],"review_version":1}